RFR: 8364235: Fix for JDK-8361447 breaks the alignment requirements for GuardedMemory
David Holmes
dholmes at openjdk.org
Tue Jul 29 07:50:55 UTC 2025
On Tue, 29 Jul 2025 06:38:59 GMT, Johan Sjölen <jsjolen at openjdk.org> wrote:
>> The fix for [JDK-8361447](https://bugs.openjdk.org/browse/JDK-8361447) added a new field to the `GuardHeader`, not realizing that the size of the `GuardHeader` must be such that the address of the user-data has the strictest necessary alignment (16-byte).
>>
>> We need to add a padding field to restore the alignment.
>>
>> A static assert is added to check the alignment.
>>
>> Testing:
>> - tiers 1-3 (in progress)
>>
>> Thanks
>
> src/hotspot/share/memory/guardedMemory.hpp line 144:
>
>> 142:
>> 143: void* padding; // Ensures 16-byte alignment
>> 144:
>
> The right thing to do is to do:
>
> ```c++
> class alignas(16) GuardHeader : Guard {
> // NO void* padding
> };
That aligns the `GuardHeader` (something already done by `malloc`) but doesn't guarantee the alignment of the `_base_addr` field which is the user-data ptr.
> src/hotspot/share/memory/guardedMemory.hpp line 164:
>
>> 162:
>> 163: static_assert(sizeof(GuardHeader) % 16 == 0, "GuardHeader must be 16-byte aligned");
>> 164:
>
> `static_assert(alignof(GuardHeader) == 16, "GuardHeader must be 16-byte aligned");`
Again `malloc` already aligned the `GuardHeader`.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/26524#discussion_r2238860572
PR Review Comment: https://git.openjdk.org/jdk/pull/26524#discussion_r2238861674
More information about the hotspot-runtime-dev
mailing list